Logistic regression in R

Video tutorial on building logistic regression models and cross-validating them in R with RStudio.
Please view in HD (cog in bottom right corner).

[j.bananagans]

GRE is a standardized exam in the US that's taken by college graduates who are looking to go into graduate school. That said, this tutorial has been incredibly helpful, and I appreciate it. Hell of a lot better than trying to make sense of the book, lol.

[kuldeepaswal]

Confusion matrix function in 45:28 gives reverse values, i.e value of sensitivity is actually the value of specificity and vice=versa, value of PPV is the value of NPV and vice=versa, i don't know in which format it takes the table but i have tried all possible ways but it always reverses the role.

Bottom line: don't totally rely on software values, try to cross validate if possible.

[kavyaramesh]

Could you explain more about the specificity and sensitivity and how to interpret it?

[utpallama]

hey bro, amazing video btw this is exactly what i needed. Quick question, is there a way to do hyperparameter tuning or model tuning in logistic regression

[petrhrobar]

Hello,
great tutorial, I just wanted to ask why are you not using average marginal effects (AME) those I easire to interprete

[petrhrobar]

Again great tutorial
Forgot to ask let’s say we get a new list of student how do we evaluate new student with our model in R

[samaneh]

Hi Hefin,
Thanks for the great explanation of the polynomial model.
I understand that you can do polynomial reg. using base R that provides the p-values as well. You used two different packages to achieve this, Is there a reason for that?
Thanks,
Sami

[brazilfootball]

Dumb question, but the way we interpret our results given different variables for our logistic regression, it's always "in the direction of whatever our "1" value is"? As in, a positive odds ratio says we are X times likely to see (fill in blank for definition of value 1 = "get admitted to college"; "catch a fish while fishing", "Win the lottery")? This is not the odds of getting a "0", correct?

[sebastianprohaska]

Hey Hefin,
first of all: THANK YOU! Your tutorials helped me a lot, please keeping up that great work!
I have a short question related to the Confusion Matrix: The last line says --> 'Positive' Class : 0. What does that mean?
Because I thought that the "Positive Class" for admit is 1? Thank you in advance for your feedback.

Kind regards
Sebastian

[anirudhmgautam]

Thank you so much for the amazing content, I have a small doubt, when I run the model, R shows an error for the confusion matrix " Error: `data` and `reference` should be factors with the same levels." any idea why this is happening or how to rectify it . I am a novice when it comes to R and any suggestion would be welcome thanks !

[glawtonmoore]

GRE = graduate record exam - a standards base testing, evaluating preparedness for grad school

[tamaraabzhandadze]

Thank you for the nice tutorial. I have used it for my project and so far i have made the nice progress. I have however one question. How can you get the result for each fold? Because based on the tutorial we can get the sum of the 10 folds, but i really need individual values for each fold.

[jayjayf]

Also u didn’t split the data into train and test data, when u do the cross evaluation does it do that for you ?

[thejll]

GRE: graduate record examination. A test you take, in the US, to enter a Masters or PhD programme. High GRE scores means you are good - the analysis results imply you are less likely to get in if your GRE is high?? That does not sound right... ah - it is the modification of the GPA that you apply, that does this!

[puruk]

how do i plot a ROC curve for the cross validation one?

[kabrantie]

Any reason why you didn't scale the data first? given GPA and other variables are different units?

[jayjayf]

Are u a data scientist as your day job? Which is better to learn python or R?

[spike]

Great video. Unfortunately I cant load the package "caret":
I get this error:
Error: package or namespace load failed for ‘caret’ in loadNamespace(j <- i[[1L]], c(lib.loc, .libPaths()), versionCheck = vI[[j]]):
there is no package called ‘gower’
In addition: Warning message:
package ‘caret’ was built under R version 3.4.4

[azizamer]

When it reaches the confusion matrix line, I get this error:
Error: `data` and `reference` should be factors with the same levels.

Can you please help to fix it?

Thank you

[nilayjaiswal]

> GPAModel <- glm(Category_Gender ~ GPA, data = myData, family = "binomial")
Warning messages:
1: glm.fit: algorithm did not converge
2: glm.fit: fitted probabilities numerically 0 or 1 occurred